Grex Oldcoop Conference

Item 407: PROPOSAL: Levels of User Access to Internet/EMail Resources

Entered by cross on Thu Apr 19 13:45:47 2007:

At the last board meeting, there was discussion of redoing the way we grant
access to different resources to our users.  Right now, we have (defacto)
three basic categories that people fall into:

1) No access.  You're a brand new user, and we give you no access to
   outbound Internet or email resources.  You can still create a web page,
   and receiving incoming email.

2) Minimal access.  You're still a newish user, but you have somehow
   been granted minimal Internet access and possibly the ability to send
   outbound email.  Minimal Internet access is defined as access to the web
   (as a client), making DNS requests, using finger, whois, and maybe talk
   (whether people actually *use* these last three is questionable).

3) All access.  You can send (almost) anything you want to the Internet
   and can send outbound email.  Currently restricted to members.

These three categories were not necessarily planned, but rather grew up in
response to changing conditions on grex and the Internet as a whole.  We've
never *really* formalized any of the categories, nor how one moves between
them.  There are some defacto rules and perhaps some old member decisions,
but we haven't evaluated this in a long time.  I submit that it is now time
to do so; in particular, whatever decisions we've made in the past really
need to be re-evaluated in light of how the electronic landscape has changed
since we made those decisions (probably in the 1990s).

What follows are my proposals for how we should handle this issue, as well
as my rationale.  This item is for discussion of this proposal and this
issue as a whole.  Please note, I am not a lawyer; I don't understand how
this would affect, say, our common carrier status for instance, but I'm
hoping that those who do will weigh in.

I propose that we maintain the following three levels of access and
associated methods for transitioning between levels.  Note that each level
is a superset of the level below it:

1) No access.  This is for new users; they are given no Internet access
   at all and are blocked from sending and receiving Internet email, though
   they can send and receive as much local email as they like (for instance,
   they can send email to staff).  This is similar to how things are
   presently set up.

   RATIONALE: The Internet has changed drastically since 1991, when Grex
   first set up shop (though I understand that the first connection to the
   net did not come for another couple of years).  In particular, the
   Internet is far more dangerous now than it was then.  We simply cannot
   allow access to the Internet, even minimal access, for untrusted users
   anymore.  We have seen too many attacks originate from grex for this to
   be practical anymore.  With respect to email, we've seen that the vast
   majority of users just don't read email on grex; there's no point in
   creating them a mailbox that quickly becomes filled with spam unless they
   ask for it, which brings us to the next level of access.

2) Minimal access.  Again, this is similar to the way things are currently
   set up; we allow minimal access to some common protocols (HTTP, DNS,
   finger, talk, etc) and allow users to send and receive Internet email,
   but otherwise constrain there access.  One transitions from the no access
   category to the minimal access category by requesting minimal access.  In
   practice, this will involve running a program from grex's command line;
   that program will send a message to a team of volunteers taken from the
   community who, if they approve of the request, will run another program
   that moves the user to the minimal access category.

   RATIONALE: This is little different from how we do things right now,
   except that it combines requests for Internet email access with requests
   for minimal Internet access, and adds community verification (a term I
   heard at the last board meeting, though I do not recall who coined it;
   perhaps cmcgee?).  The idea is that the social interaction required by
   actually requesting access from a human being will be enough to deter
   most Internet predators.  As Steve has observed, most of the people who
   will abuse grex's services want to remain under the radar and are loathe
   to do something public and personal (like request Internet access).  It's
   easier for them just to go somewhere else.  Further, we tie minimal
   network access to email since it seems that there is little reason to
   separate them; in the case where we did separate them, the procedure for
   requesting access for one would be nearly identical to requesting access
   to another.  It would put an undue burden on both our community
   volunteers and our users to separate the requests.

3) Verified access.  This is full (almost) unrestricted access.  (Almost in
   the sense that a verified user still can't, e.g., allocate a privileged
   port, for instance).  A user transitions to this level by providing some
   form of identification to grex that validates their identity.  This might
   mean either making a minimal payment via PayPal, or using some form of
   identification comparable to one of those used for, e.g., gaining
   membership.  Note that all members are automatically verified, but that
   verification does not require membership.  Once a user is verified, that
   users account becomes non-reapable for somewhere on the order of five
   years since the last time the user logged in.

   RATIONALE: It has always struck me as strange that the primary criterion
   for outbound access has been membership.  We say we want to prevent the
   situation where members are granted special privileges over non-members,
   but that's basically what ends up happening.  However, it seems clear
   that the intent is to prevent abuse, and to require some sort of
   verification before granted unrestricted access to the Internet; this is
   certainly a reasonable step to help mitigate risk.  But if that's the
   case, then let us decouple the notion of verification from the notion of
   membership.  Verified users get outbound Internet access, as do members,
   but membership is no longer the focus of verification.

   This will clean up a few edge cases.  For instance, what happens when a
   user who has previously been a member lets his or her membership lapse?
   That user has been verified and was trusted to access the Internet, why
   suddenly is the user no longer permitted to access the network?  They
   haven't been *unverified*, they're just not a member presently.  We are
   in a situation presently where some members have decided that given the
   surplus of cash we have on hand at the moment, grex does not need their
   continued financial support right now, and have decided to let their
   memberships lapse.  Very well.  But why do we take away Internet access
   from them?  We know who they are, we know we can count on them not to
   attack remote hosts.  So why not let them retain outbound access?
   Further, suppose their account becomes inactive for a time; we went to
   the trouble to verify them, so lets let them ride inactivity for a little
   while.  We may remove their outbound access until they come back and run
   some sort of a program to validate that, yes, it really is them logging
   in again, but there's no need to remove and re-validate the account after
   only three months.

This is my proposal.  What do others think?
44 responses total.

#1 of 44 by remmers on Thu Apr 19 14:59:00 2007:

I support decoupling membership from internet access.


#2 of 44 by mary on Thu Apr 19 15:05:21 2007:

I like it.


#3 of 44 by kingjon on Thu Apr 19 15:35:21 2007:

As do I.


#4 of 44 by keesan on Thu Apr 19 15:53:51 2007:

I doubt that most people who want to use a browser here also want to receive
email from outside grex.  There are probably also people who want the email
but not the other internet privileges.  I set up spam filters for people from
India, Mexico, and Indonesia who appear to just like using shell account
email.  So please decouple them.


#5 of 44 by keesan on Thu Apr 19 15:55:42 2007:

I just checked and one person (from Poland) is using pine and six are using
lynx.  This suggests that there are more people using grex to browse than for
email, and they probably do not all want the email.


#6 of 44 by cross on Thu Apr 19 17:03:56 2007:

Regarding #4; Well, just because one *can* receive Internet email doesn't
mean that one *has* to.  One can certainly forward it elsewhere or to
/dev/null or just opt-out.  And similarly, just because one *can* use a web
browser or other minimal form of outgoing access doesn't mean that one has
to.  On the other hand, if we decouple the request process, that does place
an additional burden on the volunteers (these are not necessarily staff
members) to authorize the same account for multiple services.  It seems like
an unnecessary complication to separate the two.  Sindi, what's your
reasoning?  It imposes no additional burden on the user to have something that
he or she doesn't use....


#7 of 44 by keesan on Thu Apr 19 21:56:29 2007:

It fills up disk space for no reason, and probably puts a bit more strain on
the mail receiving software and any future antispam software.  Why not just
have them request internet, or mail, or both?


#8 of 44 by kingjon on Thu Apr 19 23:55:29 2007:

#7: That's what I thought he was saying -- just a user wouldn't have to file
two separate requests for email and Internet access.


#9 of 44 by cross on Fri Apr 20 00:00:23 2007:

Regarding #8; That is what I'm saying.


#10 of 44 by krj on Fri Apr 20 14:26:40 2007:

Regarding allowing all users to create web pages:  M-net, in the 
policy conference, has reports that malicious users have been using
M-net to host phishing pages, and M-net's hosting provider 
is threatening to cut access if this happens again.

Has this been an issue for Grex?  Can Grex move fast if it does 
become an issue?
 
(A phishing page is the page set up to collect financial information,
such as bank account information or PayPal passwords, on a fraudulent
basis.)


#11 of 44 by cross on Fri Apr 20 14:34:56 2007:

I hadn't really addressed the web page issue.  It has been an issue in the
past, but surprisingly infrequently.

But now that you mention it, it strikes me that no access users probably
should *not* have the ability to throw up a web page.  However, limited access
users probably should; would `community verification' be sufficient to insure
no abuse (or at least `managable' abuse) from otherwise non-validated users?


#12 of 44 by krj on Fri Apr 20 17:28:09 2007:

I hate to pull a function before it is necessary for Grex to 
pull it.  Here's a suggestion:
 
If Grex is not being threatened due to phishing pages, allow 
no-access users to create web pages, as we do today.  Obviously
staff continues to swat down phishing pages as they are reported.

If such a threat arises, then have the hooks in place so that 
web page creation can be switched to the limited-access category
at a moment's notice.   This might mean board pre-approving the 
switch in advance, at staff's judgement.


#13 of 44 by tod on Fri Apr 20 17:41:51 2007:

Why not add a server side include banner on the webpages?


#14 of 44 by keesan on Fri Apr 20 17:45:08 2007:

I would volunteer to preapprove any webpages to be posted by non-members, and
any changes to them made later.  


#15 of 44 by cross on Fri Apr 20 17:48:05 2007:

Regarding #12; I'm fine with that.

Regarding #14; One of the purposes of this proposal is to reduce the
privileges of `members.'  Membership rights should be limited to the
governance of the corporation, not access privileges.  I think what you meant
was, `webpages posted by non-verified users.'  But I think that we really
don't want to get into the business of `approving' web pages or anything else.


#16 of 44 by remmers on Fri Apr 20 18:16:55 2007:

Re #13:  Couldn't JavaScript be used to override a server side banner?

Re #15 re #14:  I agree.


#17 of 44 by tod on Fri Apr 20 20:09:37 2007:

re #16
 Couldn't JavaScript be used to override a server side banner?

That's a good question and I'm not sure if it's been done.  I'm not thinking
of pop-ups but actual includes which would be at the top of every webpage.


#18 of 44 by nharmon on Fri Apr 20 20:17:48 2007:

Javascript could be used to override the server side banner. Maybe with
a InnerHTML property.


#19 of 44 by tod on Fri Apr 20 20:49:48 2007:

So you're talking about something that would cover the HTML of the banner?
I'm thinking the httpd config could include a shtml at the beginning of all
userpages and it would include banner graphic and text plus the TAGS to show
it is Grex, etc..


#20 of 44 by slynne on Fri Apr 20 20:56:15 2007:

well, if there were a way to have a server side banner, that would be
the best solution imho. 


#21 of 44 by nharmon on Fri Apr 20 20:57:36 2007:

Javascript allows you to dynamically change the HTML.


#22 of 44 by mcnally on Fri Apr 20 21:47:29 2007:

 At the very least, Grex should have some language in its terms of use
 that empowers staff to take web pages off-line in response to complaints
 about things like phishing.  It should be a clearly defined system policy
 and without authorization under such a policy staff shouldn't remove
 anything -- otherwise censoring one web site and not another could be
 argued to be implicit approval or endorsement of the other.


#23 of 44 by cross on Fri Apr 20 22:55:35 2007:

Well, perhaps we need more data on the frequency of phishing attempts and the
consequences we've been threatened with; I honestly am out of the loop on that
right now.  I agree 100% with #22, though.


#24 of 44 by maus on Sun Apr 22 00:42:48 2007:

I like the idea of decoupling membership from service, and I think that
the tiered access system makes sense, where increased privilege is
granted by increased exposure to the staff, thereby increasing personal
accountability. 

keesan, the idea of having to have webpages vetted concerns me, and I
suspect it would concern others as well. I would prefer that staff had
the right to revoke before giving them the responsibility to vet a page
beforehand. Requiring approval before putting up content without some
reason to suspect the user in question would have a chilling effect.
That said, I also think staff should have a mechanism in place to be
able to yank a webpage that violates TOS. I also think it would be a
good idea to have the TOS looked over, reaffirmed by staff and reposted
in a prominent place when we have the next system upgrade. 


#25 of 44 by gelinas on Sun Apr 22 03:57:55 2007:

From the MotD:

} To see statements of grex principles and limits, run
} 
}         /usr/local/bin/grex-principles
}         /usr/local/bin/grex-limits

Thus, you can use !grex-limits at the next available prompt.  Comments, of
course, are welcome.

I like the idea of decoupling membership and access levels.  I do not like
the idea of coupling access levels and donations.  Verification, and thus
increased access, should NOT rely on a contribution to the corporation.
I don't want to get into fees for services rendered.


#26 of 44 by mcnally on Sun Apr 22 04:00:56 2007:

 Boy, those limits haven't been updated in a while, have they?
 They don't appear to say anything about phishing.  Is that in
 the principles document?


#27 of 44 by cross on Sun Apr 22 16:19:13 2007:

Regarding #25; No, I wouldn't think that verification would be tied to
contributions to the corporation; my PayPal example was just one method by
which one could be verified.  Mailing a photocopy of an ID to the treasurer
or some other designated entity would be another method (though, I guess, then
one is donating a stamp to the corporation.  :-))


#28 of 44 by krokus on Mon Apr 23 02:54:22 2007:

Part of the updating could be using terminology that others would
recognize ToS.  Those filenames would make me think of what the ideals
are, and the limitations of the system.


#29 of 44 by maus on Mon Apr 23 06:12:45 2007:

While those two are a good starting point, I would recommend that they
be extended in such a way as to provide a means for enforcement.
Additionally, phrasing it as "it would be nice if" decreases
enforceability. Because the rules are fairly specific, they are harder
to enforce on the edge cases ("well, it doesn't say I can't use my
account to trick people out of their credit card information"). Lastly,
Cyberspace Communications needs to provide a means by which liability
can be transfered to the infringing party; that is, if I use Grex to do
something illegal, Cyberspace Communications needs to make sure that
they have already established legal grounds by which they can sue my
sorry tail if they get sued. 

Oh, and the terms of service also need cheese. 


#30 of 44 by aruba on Wed Apr 25 17:56:23 2007:

I am a little confused why Dan says we've never formalized the internet 
access categories he describes in #0; the membership category was 
formalized a long time ago, and the distinction between the first two was 
formalized at the last board meeting.

The content of Dan's proposal seems to be that we should allow people full 
internet access if they are verified but not members.  This was in fact the 
original intent when Grex first instituted an internet access policy, but 
for various reasons full access has always been linked to membership.

I am in favor, in principle, of allowing verified members full internet 
access.  There are a few logistical problems we need to consider, however.

1. How long should a validation be valid?  If I take out a Paypal account 
today, verify it, pay Grex a dollar, then move next year but don't tell 
Grex, the validation information from Paypal is not very valid.  If I did 
something destructive and Grex handed over my info to law enforcement, I 
doubt that they could find me.

Now, granted, this is a bit far-fetched; it's unlikely someone will go to 
the trouble to gain privileges on Grex in order to abuse them several years 
in the future.  But it is certainly possible.

If we allow verification to last 5 years, I envision us having a lot of 
accounts on the verified rolls which have not been logged into for a long 
time.

2. Someone needs to accept and record validation information.  I assume 
that will be the treasurer (currently me).  I don't expect a deluge of 
people requesting validation, so it will probably be fine; but I am aware 
that everything that gets added to the treasurer's job will make it harder 
to find someone to do it in the future.  Also, presumably, when a verified 
account's verification period is up, presumably someone will need to remind 
(nag) that person to re-validate.  I presume that will also fall to the 
treasurer.  If it's been 5 years, it's not at all unlikely that it will be 
hard to find an email that works.

3. Do we accept the same forms of ID that we always have?  Currently that 
includes school IDs and library cards.  (See ~aruba/idpolicy for a 
description of currently acceptable IDs.)  No one has used such an ID in a 
while, I must say; by far the most popular form of ID these days is a 
verified Paypal membership.  I bring the topic up because if we are 
reducing the bar required for access, perhaps we should make up for that by 
requiring a little more ID.  (In other words, it's possible that the 
necessity of sending $6 along with one's library card may have discouraged 
certain vandals who will not be discouraged by simply sending the library 
card.)

People should be aware that we will probably lose a few members as a result 
of changing the policy, because there have lways been a (changing) handful 
of people who become members in order to have the internet privileges.  I'd 
estimate there are 1-5 such members at any given time.  We can afford to 
lose that much income.


#31 of 44 by nharmon on Wed Apr 25 19:07:56 2007:

I generally become a member in order to have the internet privileges and
to vote. Although I would also become a member if Grex needed the money,
but that hasn't been the case lately.


#32 of 44 by cross on Wed Apr 25 20:01:36 2007:

Regarding #30; (First Para): Sorry, that was ambiguous; I was trying to give
a brief recap of the discussion at the board meeting, but I didn't make that
clear.  At the board meeting, we said that the access levels had never really
been formalized, and then sorta formalized them, but also said we should take
it to the membership (unless I misunderstood things).  Hence the proposal.

First point: My vision was that verification should take place, and then
remain valid for as long as the account is active, plus a grace period
aftwards.  I'm okay with the grace period being as short as a year or two,
five is good too, but that's a detail that will have to be formalized.

Second point: I'm not sure we need to make verification a responsiblity of
the treasurer.  If most of it is done by Paypal, then we could write a script
to do it automagically (just query PayPal as necessary, and update the user's
primary group ID).  I guess that other mechanisms need a bit of thought, if
for no other reason than that we are constrained by who actually picks up the
(physical) mail at the PO Box.

Third point: Personally, I don't think that library cards are sufficient
anymore.  Some sort of picture ID is probably best.


#33 of 44 by aruba on Thu Apr 26 05:44:15 2007:

Any automatic querying of Paypal would require storing Grex's Paypal
password somewhere; I'm not at all crazy about that idea.


#34 of 44 by ric on Sat May 5 04:07:47 2007:

That's probably not true.

Paypal's integration does NOT require you to store your paypal account
information in scripts.  Merchants direct buyers to their paypal store, and
paypal redirects them via a special link back, or possibly makes an alternate
request, either way the informatio nca be verified WITHOUT needing the grex
paypal accont password *BY* the receiving script.


#35 of 44 by aruba on Thu May 10 04:01:32 2007:

Re #34: Dan was talking about automatically querying Paypal from time to
time to get info on recent transactions; to do that, you need the password.

However, Paypal does send a confirmation notice when someone sends money to
Grex.  I suppose those notices could be (in theory) parsed when they come
in, and people added to the appropriate groups automatically.  We'd have to
verify that the messages really came from Paypal, though, or someone could
send such a message themselves and become falsely validated.


#36 of 44 by cross on Thu May 10 14:11:59 2007:

Hmm; I wonder if there's a way to ask them to send a signed message or
something....


#37 of 44 by maus on Thu May 10 19:08:54 2007:

I think cryptographic signing, creation and management of keys, etc
would be far too complicated for the majority of our users; keesan would
give birth to kittens and complain loudly that it is not part of default
pine, and damnit, telnetting in and using pine right on the server is
the way the gods meant for humankind to do email, thankyouverymuch. 


#38 of 44 by cross on Thu May 10 19:52:58 2007:

Well, only between PayPal and the script that interprets whatever comes back
from PayPal, not for the end users.


#39 of 44 by maus on Fri May 11 22:53:59 2007:

Oh, have PayPal send a signed message. I thought you meant have the
prospective validated member send a signed message. Pardon my confusion.


#40 of 44 by cross on Fri May 11 23:00:18 2007:

Not at all!


#41 of 44 by eteepell on Sat May 12 19:40:28 2007:

What has been proposed seems quite appropriate and reasonable, without any
particular change. 

Having newusers with the ability to host pages
automatically always seemed unnecessary to me. Losing that function for 
new accounts does not seem to be a big deal. And probably may be a good 
idea in the long run. No reason they could not host a locally 
accessible page, but otherwise I see no reason why we cannot drop 
hosting for --brand new users--. (note emphasis)

One thing I always liked about grex
was even as a newuser there was full shell access (albeit without outbound
internet connectivity). I am seeing nothing here that would change that. 
Kudos.
Also the ability to self-create accounts on the system for INSTANT access 
is as far as I can see without any other equivalent out there.

Insofar as paying for access, I would not have any problem with any 
particular users paying for ::
--larger disk quota
--larger mailbox quota
..etc.etc.

If some people pay for more space, etc. that allows purchase of larger 
disks, better hardware, and it can then purposefully flow down to all 
users on the system as everyones space and service increases in benefit. 
Any thoughts?


#42 of 44 by eteepell on Sat May 12 19:45:10 2007:

Almost forgot, I'm not for nitpicking the levels, everyone gets email at a
certain level. If they dont use it fine, they can "> /dev/null" or fill up
their disk space quota at their option. Everyone gets specific access at
specific levels and they choose if, and when they use the access, if they ever
do. thx. (I personally dont use the email, glad it's there though. I dont get
ANY email, spam or otherwise, just FYI).


#43 of 44 by kingjon on Sat May 12 19:49:29 2007:

My only objection to adding web hosting to the list of privileges restricted to
activated users is that Grex was founded on the principles of free speech, and
if the verification process required a name that could be harmed.

On the other hand, Grex was also designed as an online community, not as a
fee-for-service model, on which grounds I would object to letting users pay for
more disk space, etc. 

(The founding was before my time, of course, so I know whereof I speak only
second-hand.)


#44 of 44 by cross on Sun May 13 15:18:32 2007:

I don't think we need to ask people to pay for more disk space, mail quota,
etc.

The thing about web pages is that we need to balance out the risk of an
unverified user creating a phishing hole (get it?  ha ha ha) versus legitimate
users who want to create pseudo-anonymous web sites.  I'm honestly not sure
where the balance should be there.


There are no more items selected.

You have several choices: